For everything wrong in the world, heart disease remains the number one killer. Up to 20% of people die alone from sudden cardiac death (SCD). SCD is not a heart attack that occurs when arteries are blocked, but an electrical problem that prevents the heart from beating properly.
SCD can strike seemingly out of nowhere. But a new AI developed by John Hopkins professor Natalia Trayanova, led by research scientist Dan Popescu (a former graduate student under Trayanova), can detect SCD long before it happens. A decade ago, in fact. The results were published in a recent issue of nature cardiovascular research.
You might read such a prediction as a minority report death sentence. Who wants to know what they will die of? But as Trayanova explains, predictive tools like this AI can transform the way doctors treat their patients. And ultimately, they will help prevent deaths that humans could not predict before.
If you are predisposed to SCD today, doctors treat it by implanting a defibrillator. But knowing if someone is predisposed to SCD has been problematic. The main way to determine it now is to look at the total blood flow from the heart’s chambers per beat, known as the ejection fraction. For example, if you have an ejection fraction of 50%, that means that each beat pumps out only 50% of the blood in a filled ventricle.
“When a patient has less than 30% ejection fraction [doctors] add a defibrillator; if not, then not,” says Trayanova. “That’s it.”
With 20% of people still dying from SCD annually, it is clear that many defibrillator candidates are being overlooked. On the flip side, however, Trayanova points out that this 30% ejection fraction rule of thumb also results in many people getting defibrillators they don’t actually need. (One study found 23% of people who received these devices didn’t need them — with that number skyrocketing to 40% in some healthcare facilities.)
“It’s not fun to live with this device,” says Trayanova. “It can discharge, components can fail, and there are studies showing it increases mortality. It’s like a horse kicking you in the chest – it’s so painful.”
Trayanova’s lab built its AI to close the large prognostic gaps in SCD. To construct the system, her team trained a machine using 10 years of medical records from 156 people with heart disease who agreed to share their medical information. They shared everything in their medical records, from MRIs of their hearts to 22 other potentially relevant pieces of information, including race, weight, drug use and high blood pressure.
By feeding all MRI scans into a machine learning system, the researchers were able to identify hidden patterns, such as: B. How scar tissue and other components of a person’s heart predispose them to SCD. (Technically, a second AI was built to understand how smoking or other factors can also affect this probability.) After creating their software, the researchers validated their tool using patient data from 60 health centers across the US. The AI outperformed doctors in performing diagnoses.
Trayanova’s AI model can provide a personalized estimate of the presence of SCD in a given year over a 10-year period. This allows a doctor to discuss the best course of action with their patient. The AI also reveals its own certainty to the doctor. “You can say, ‘Okay, you have a 50 percent chance of having [an SCD] in five years, but I’m so sure about it [estimate],’” says Trayanova.
For someone flagged as likely to need a defibrillator, that’s good because they’ve been properly diagnosed and their life could be saved. For someone who has poor ejection fraction but is otherwise at low risk, according to AI, a doctor might suggest follow-up visits, medication, and/or lifestyle changes to help their heart without surgery. Maybe this patient still needs an implanted defibrillator, but he could also avoid the discomfort thanks to this AI.
While Trayanova is not a practicing physician herself, she paints a compelling picture of how seemingly cold, calculating AI tools can disrupt imperfect medical best practices and ultimately serve as a mechanism physicians can use to provide long-term care for their patients. The bottleneck now is getting these AI tools into the hands of more doctors, as Trayanova only works with a limited pool of clinicians through her research lab.
Ideally, these AIs would be integrated directly into patient record platforms so that any doctor can access them at any time. In fact, the Mayo Clinic has Millions of dollars spent integrating AI into its heart disease treatment systems; but this AI applies only to ECG measurements. The US healthcare system just isn’t well structured to quickly integrate cutting-edge AI models into its software.
“There are many options for moving forward [with AI], but it cannot be done by a single researcher,” says Trayanova. “There has to be a lot of buy-in from the site [medical] community.”
https://www.fastcompany.com/90740876/this-ai-can-prevent-your-death-10-years-from-now-so-how-does-that-work?partner=feedburner&utm_source=feedburner&utm_medium=feed&utm_campaign=feedburner+fastcompany&utm_content=feedburner This AI can prevent your death in 10 years